Thank you Anderson,
For transnational motion parameters we have the estimates in x, y, z. Same thing for rotational ==> I have six columns of motion parameters and I have two choices:
1. Feed the six columns to GLM
2. Calculate the sum square of x,y,z for every DTI direction==> two columns of motion parameters one for transnational and the other on for rotational.
I am not sure which choice will fit better for GLM, including 6 columns or just two columns (sum squares)?
Thank you for any advice.
Jon
Hi Jon,
You know the answers for these questions...:
1) As far as the GLM is concerned yes, you can put these or any other nuisance variable. If it will absorb some or all the motion variability, then it's another story for you to investigate.
2) Depends on the design and contrasts. Please see Jeanette Mumford's page on this matter: http://mumford.fmripower.org/mean_centering/
All the best,
Anderson
On 18 November 2016 at 10:44, John anderson <[log in to unmask]> wrote:
Dear FSL experts,
I have DWI data for two groups. The images are acquired from children 10-15 year old. I am worried about motion in this data set. After applying eddy correction tool (ecc) on the raw DWI data. I got the file "data.ecc.log" as an output of applying ecc tool . Then I calulated the motion parameters "transnational, rotational" for every DTI direction
My questions are:
1. Can I include these motion parameters (transnational and rotational) a covariate in GLM.
2. Do I need to demean these acquisitions.
I highly appreciate any suggestion or advice
Many thanks
Jon
|